TypeSafe AI’s Jev has become the fastest-adopted artificial intelligence model in Vercel AI Gateway’s history, reaching 35.3% of paid teams within two weeks of its September 2026 launch. The decision-focused model achieved nearly twice the adoption rate of Anthropic’s Claude Sonnet 5.5, which reached 18.1% of paid teams during its corresponding launch period.
According to Vercel’s October 2026 AI Gateway Production Index, published October 15, Jev also received more requests than any other model launched in September. The findings are based on activity through Vercel’s platform rather than the entire AI model market.
How Jev Differs From Traditional AI Models
Released in limited availability on September 15, Jev is designed to make decisions from predefined options instead of generating long, open-ended responses. This approach allows the model to handle tasks such as data classification, content filtering, and selecting actions within automated workflows.
By restricting the range of possible outputs, Jev can process requests using fewer computational resources than general-purpose language models. This makes it suitable for applications that must classify large amounts of information quickly and at lower cost.
The model has attracted interest from developers seeking specialized AI systems for tasks that do not require extensive text generation. OpenAI and Amazon have also introduced competing decision-oriented models following Jev’s release.
Vercel was among Jev’s early adopters, using it to replace OpenAI’s GPT-6 Luna for identifying potentially harmful commands. The company operates AI Gateway, a unified application programming interface that allows developers to access AI models from multiple providers.
Jev Records Fast Adoption With Low Processing Costs
Vercel reported that Jev reached 35.3% of paid teams during its first two weeks, compared with 18.1% for Sonnet 5.5 and 14.9% for GPT-6 Luna. Among teams that tried Jev during its first week, 55.6% returned to use it again in the second week.
Despite its rapid adoption, Jev accounted for only 0.27% of gateway spending during its initial two-week period. Across September, the model handled approximately 1% of processed tokens while representing just 0.15% of total spending.
The difference between adoption and expenditure reflects Jev’s relatively lightweight processing requirements. Its usage figures indicate that developers are incorporating specialized models into production applications without necessarily generating substantial inference costs.
Open-Weight AI Models Gain Usage Across Vercel
Vercel’s report also found that open-weight AI models accounted for 71.5% of tokens processed through its gateway in September. These models represented approximately 23% of total spending, indicating that lower-cost alternatives are handling a substantial share of AI workloads.
Meanwhile, spending on frontier models increased tenfold between December 2025 and September 2026, even as their percentage of total gateway spending declined. Overall AI Gateway expenditure increased approximately thirteenfold during that period, while the average price per token fell 47%.
The figures suggest developers are increasingly combining specialized, open-weight, and frontier AI models for different applications. However, Vercel’s findings do not capture all industry activity, particularly usage through providers’ direct APIs.
Featured image credits: Magnific.com
For more stories like it, click the +Follow button at the top of this page to follow us.